Table 4.
Accuracy of NLP results by mammographic finding
| Gold standard from manual review | Accuracy | Sensitivity | Specificity | PPV | NPV | |||
|---|---|---|---|---|---|---|---|---|
| Results from NLP algorithm | Positive | Negative | ||||||
| Mass | positive | 5 | 1 | 0.99 | 1.00 | 0.99 | 0.83 | 1.00 |
| negative | 0 | 94 | ||||||
| Calcification | positive | 17 | 0 | 0.99 | 0.94 | 1.00 | 1.00 | 0.99 |
| negative | 1 | 82 | ||||||
| Asymmetry | positive | 11 | 3 | 0.96 | 0.92 | 0.97 | 0.79 | 0.99 |
| negative | 1 | 85 | ||||||
| Architectural distortion | positive | 2 | 2 | 0.98 | 1.00 | 0.98 | 0.50 | 1.00 |
| negative | 0 | 96 | ||||||
Note: PPV, positive predictive value; NPV, negative predictive value